senioirity of sovereign debts

8/9/2019 Senioirity of Sovereign Debts

http://slidepdf.com/reader/full/senioirity-of-sovereign-debts 1/40

249

7Seniority of Sovereign Debts

A formal insolvency regime typically sets out, in general terms, how dif-ferent types of claims on a distressed private firm will be treated in a re-structuring and the order of payment in the event of outright liquidation.Bankruptcy law usually indicates that equity is junior to debt: Debt gets

paid first, and if debt cannot be paid, debtholders may take control of thefirm. Back taxes get paid ahead of most private debts. Different kinds of debts may have different levels of priority—for example, debts backed bycollateral are treated better than unsecured debts. These rules tell a firm’screditors where their claims stand in the pecking order.

In contrast, no formal rules of priority lay out how different types of claims on a distressed sovereign will be treated. Nor does a court have thepower to force a sovereign government to respect any rules of priority.1

The difficulties of taking effective legal action against a sovereign—dis-

cussed in depth in chapter 8—can give the sovereign the ability to dis-criminate in favor of some classes of claims. Different creditors will arguethat their claims should be given priority. As a result, the relative stand-ing of various claims on a sovereign government is often ambiguous. Thepriority that the sovereign opts to grant different types of debt may notmatch creditors’ prior expectations.

1. In some countries, certain public entities are subject to an insolvency regime. Municipal-ities in the United States, for example, are subject to Chapter 9 of the US bankruptcy code.

Municipal bankruptcy regimes may even explicitly declare that public employee wages andsome public expenditures are senior to financial liabilities. US states are not subject to an in-solvency regime. However, some state laws may set out the relative seniority of variousclaims. Here, too, some spending items linked to the provision of public services may haveformal priority over debt claims.



Priority can be absolute or relative. Absolute priority means that claimsat the top get paid in full, those in the middle get paid in part, and thoseat the bottom get nothing. Relative priority assures only those at the top of the priority structure better treatment than those at the bottom. Nobodymay get paid in full. In corporate bankruptcy, the rules of absolute prior-

ity—the ranking of claims in the event of liquidation—determines the bar-gaining power of different participants during the restructuring negotia-tions, and thus drive the relative priority accorded to different claims. Nocomparable rules of absolute priority cast a shadow over a sovereign re-structuring. Since a sovereign in default usually is unable to grant ab-solute priority to many—if any—creditors, the debate is usually over rel-ative priority.

Most sovereigns do respect a number of informal rules, avoiding totalchaos. The priority traditionally granted to creditors like the IMF, the

World Bank, and other multilateral development banks (MDBs) is almostalways respected, in part because these international financial institutions(IFIs) usually refinance their maturing debt rather than demand full pay-ment after a default. In some cases the IMF may go beyond simply re-financing its existing debt and provide new debtor-in-possession (DIP) fi-nancing during a crisis.2 Different external-law bonds are usually treatedroughly equally: The structure and payment mechanics of a typical inter-national bond make it hard for a sovereign to restructure one bond issuewhile sparing another. However, these informal rules still leave plenty of

sources of contention. Should the sovereign’s debt to private externalcreditors be treated better, or worse, than its debt to other governments?Should the sovereign’s external bonded debt be treated differently thanits external bank loans? Should foreign-currency debts held by domesticcreditors be treated better or worse than, or the same as, those held byexternal creditors? Should debt held by domestic banks be treated on dif-ferent terms in a restructuring? Should the IOUs from pension and wagearrears be treated better, or worse, than other domestic debts?

This chapter examines the current debate about the relative treatment

of various types of sovereign debt—a debate that can be interpreted as ar-guments about the priority structure that a bankrupt sovereign shouldput in place in the absence of any binding, formal rules of priority. It isdivided into three broad sections. The first defends both the informal pri-ority currently given to IFI debt and the principle that Paris Club creditorsshould condition their restructuring on a “comparable” restructuring of debts owed to private creditors.

The second section discusses the most difficult and controversial issuein most sovereign restructurings: the relative treatment of domestic and

external debt.

250 BAILOUTS OR BAIL-INS?

2. The term “debtor-in-possession (DIP) financing” comes from Chapter 11 of the US bank-ruptcy code.



The third section examines proposals to introduce a more formal prior-ity structure into the sovereign debt market, whether to limit the risk of debt dilution or to make debt restructurings more orderly. We concludethat it is not possible to develop an enforceable system that substantiallyimproves on the status quo.

Our focus on the issue of priority reflects our sense that the debate onsovereign debt restructuring has been framed too narrowly. Too much at-tention has been placed on the potential of the shift from syndicated bankloans to international bonds to give rise to new forms of collective actionproblems. Too little attention has been placed on the far bigger issue of de-ciding who is in and who is out of the restructuring. It is worth noting thatproposals to create an international bankruptcy mechanism for sover-eigns—discussed in depth in chapter 8—would not necessarily clarify therelative seniority of different claims on a sovereign. The IMF’s proposal,

for example, did not cover domestic debt and allowed a sovereign toleave some unsecured external debt out of its restructuring.

We have little doubt that difficulties in reaching agreement on the pri-ority that should be given to different classes of debt are likely to be a farmore important impediment to the rapid resolution of a severe sovereigndebt crisis than holdout litigation. But we also believe that the obstaclesto creating a formal priority structure—and then forcing a sovereign toabide by it—are almost certainly too large to overcome. The real challengeis finding ways to make the current ad hoc and informal priority regime

work better. Here, as elsewhere, we believe that incremental steps—moreflexible Paris Club restructuring terms for middle-income countries andgreater understanding of the case for restructuring domestic and externaldebt on different terms—is more likely to tangibly improve the debt re-structuring “architecture” than more radical reforms.

Relative Treatment of Different Sovereign Claims

A sovereign “bankruptcy” is distinguished by the diversity of differentclaims on the sovereign’s resources as well as by the absence of any au-thority that has the power to supervise the restructuring. The various po-tential claims on a distressed sovereign are worth reviewing in detail.

Direct Sovereign Debt Obligations, External and Domestic. Externalcreditors often include private-sector creditors (banks, bondholders, andothers), other governments (Paris Club members and others), and multi-lateral creditors (such as the IFIs). Domestic banks and pension funds are

often important domestic creditors. Debts owed to multilateral creditorstraditionally have been given priority. Some form of collateral may back—either fully or partially—some external debt, but most external debt islikely to be unsecured.

SENIORITY OF SOVEREIGN DEBTS 251



Contingent Liabilities. The banking system typically is the most signifi-cant source of such implicit liabilities. This is true even in the absence of asystem of deposit insurance formally backing the banking system, as thegovernment typically guarantees most bank liabilities in a crisis. The im-plicit liabilities deriving from pay-as-you-go social security systems can

also be important. Retirees promised a retirement benefit in return for thesocial security taxes they paid while working have a clear claim on the sov-ereign’s future tax resources.

Arrears on Wages, Pensions, and Social Payments. A sovereign oftenstops paying its workers and pensioners in full before falling into broad- based default. Ecuador, for example, stopped paying teachers and police-men before it stopped paying its external debt, and Argentina startedpaying workers with quasi-currencies before it stopped making debt pay-

ments. After the default, those who were not paid in full before the defaultusually want the money that they are owed.

All these financial claims compete with one another and with other“stakeholders” for a share of a sovereign government’s ongoing revenues.Creditors cannot force a sovereign that has defaulted to cut domesticspending, to raise taxes, or to devote all of its revenues to service its debts.Consequently, the treatment that a sovereign should grant to debt servicerelative to other spending priorities—defense, police, education, and pen-

sions—is among its key decisions following a default.The diversity of claims on the sovereign makes deciding who gets paidand who does not, and who takes a large haircut and who takes a smallerhaircut, intrinsically difficult. Still, a bankrupt sovereign does need toreach agreement with all those holding financial claims to end its crisis,and the terms of these agreements will create—at least ex post—a de factopriority structure.

International Financial Institutions

A sovereign typically pays debts owed to the IFIs even if it is not servic-ing its other external debts. The effective seniority of IMF and MDB claimsis not a matter of legal right. Formally, IMF and World Bank loans are justlike any of the sovereign’s other unsecured debt. However, sovereignstypically have paid the IFIs even when they are not paying other unse-cured creditors, giving the IFIs effective “preferred creditor status.” Bilat-eral official creditors have always respected the IFIs’ preferred position:Indeed, the historical willingness of bilateral creditors to restructure their

claims in order to assure payment to the IFIs has been central to the con-ception of the IFIs’ preferred status. A sovereign’s desire to maintain itsfuture access to emergency financing and a good working relationshipwith the other governments that provide the IMF with its financing pro-




vides a powerful incentive to follow the convention of paying the IMFeven if the sovereign defaults on its other debts.

A number of private investors, however, are questioning the “pre-ferred” status of the IFIs. Argentina, for example, owes a substantial sumto the IMF and the MDBs, and private creditors ask why all unsecured ex-

ternal creditors, including the IMF and the World Bank, shouldn’t take ahaircut when the sovereign cannot pay.3 Some argue that the risk of losseswould discipline the IMF and prevent it from making large loans to riskycountries.

These arguments should be rejected. The IMF is not as a private lenderseeking profitable lending opportunities, but a public institution respon-sible for stabilizing the international financial system. The IMF’s preferredstatus is central to it ability to perform its two key roles in the interna-tional financial system: It acts as a proxy for a true lender of last resort,

lending significant sums to countries facing a temporary liquidity prob-lem, and as a proxy for a sovereign bankruptcy court by providing newfinancing to a sovereign undergoing a restructuring. In both cases its fi-nancing is tied to an agreement with the crisis country’s government on aframework for policy changes, which effectively lays out the country’splan for emerging from distress.

The IMF’s preferred status lets it lend when private creditors will not,even though it has to guard against losing the taxpayers’ funds it man-ages. A classic domestic lender of last resort lends against collateral. A

sovereign government, in contrast, typically has limited ability to offercollateral: Its core asset is some combination of its commitment to futurefiscal adjustment and its commitment to buy foreign exchange that therest of the economy will generate in the future. The priority granted to theIMF provides it with the security needed to lend large quantities at rea-sonable rates against such intangible assets.

Without effective seniority, the IMF would have to act more like a pri-vate lender and pay more attention to maintaining a diverse portfolio, se-verely crimping its ability to lend large sums to major emerging econo-

mies.4 It would need to lend to a sovereign at high market rates in orderto avoid systematically losing money5 and would be less able to put


3. Argentina owes about $40 billion to the IMF, the World Bank, and the Inter-American De-velopment Bank. It owes between $80 billion and $90 billion to international bondholders,

but domestic investors hold a large share of Argentina’s international bonds. Internationalinvestors truly hold perhaps only $40 billion to $50 billion. The complexities of Argentina’srestructuring are discussed in detail later.

4. No private lender would be willing to have its exposure exclusively in relatively high-riskdebtors such as Turkey, Brazil, Argentina, and Uruguay without some assurance that its

debts would be given priority. Five countries currently account for 86 percent of the IMF’snonconcessional lending.

5. The IMF’s Articles of Agreement require it to lend on terms that do not put its resourcesat risk.



money in when other creditors are pulling money out. Since we believe,as discussed in chapter 6, that there is a need for an international sourceof emergency liquidity—appropriately priced and linked to appropriatepolicy adjustments—we also believe that payments to the IMF need tocontinue to be given priority over payments to other external creditors.

The IMF performs a second role in the international financial system.During a sovereign debt restructuring, an IMF program acts as a surro-gate—sometimes an imperfect surrogate—for the supervision that a bankruptcy judge provides during a corporate reorganization. Sovereign bankruptcy certainly is very different from corporate bankruptcy, andthe IMF’s role also clearly differs in many ways from that of a bankruptcy judge. The IMF has leverage because it lends, not because it can put anend to a “reorganization” process and either push the country into liqui-dation or force a change in the country’s “management.” Consequently,

the IMF will have, unlike a bankruptcy judge, its own money on the linein the crisis country. If the IMF did not have money—or the ability to putadditional money—on the line, then it would not have had the ability toinfluence the policies adopted by a sovereign government in default.Classically, senior lenders care the least about a borrower’s policies: Thelender’s seniority assures it payment no matter what (Gelpern 2004). TheIMF, in contrast, should use the leverage created by its preferred status toperform its public policy role of laying out a framework for the country’smacroeconomic policies after a default.

The economic argument for the preferred status of IMF lending to acountry undergoing a debt reorganization therefore parallels in part theeconomic argument for giving seniority to “new money” in bankruptcy.Granting seniority to new financing is necessary if a distressed firm isto remain viable during its reorganization. So long as the firm is worthmore as an ongoing entity than if it were to be liquidated, such seniorityserves the interest of existing creditors. IMF financing also can help to re-duce the fall in output following default, helping other creditors.6 But the

analogy to DIP financing actually understates the case for providing theIMF with seniority. IMF financing is important because access to newmoney not only can avoid larger falls in output but also provides thehook that lets the international community shape the policies of a sover-eign government in distress or default.7


6. See IMF (January 2003b) for a discussion of the role IMF financing and an IMF programcould play in a sovereign debt restructuring. It is striking that the IMF has not sought to playthis role in Argentina.

7. The argument for the seniority of MDB lending is more complicated, since most MDBloans are not provided during financial distress. The general argument for MDB seniorityrests on the need to find a way to finance, at low cost, the development of very poor coun-tries. The MDBs can only lend to risky countries at low rates, for long terms, and on an un-secured basis because of the priority that is traditionally given to the repayment of this debt.



Eliminating the IMF’s preferred creditor status would do far more harmthan good, but private creditors are right to note that the IMF has not al-ways played its role in the international financial system well. Argentinais a case in point. In 2001, the IMF provided emergency liquidity supportto a country that, in our judgment, was too indebted to have a reasonable

chance of avoiding a restructuring. Argentina’s need to repay, over time,the additional IFI debt it took on in the course of 2001 no doubt reduces,at the margins, the amount that the country’s other external creditors canexpect to recover in a restructuring. Of course, successful rescues alsohelp a country’s long-term creditors. Without its preferred status, the IMFwould not have lent as much to Argentina. But it also would not have lentas much to Brazil, Turkey, and Uruguay.

In Argentina, the IMF also has had difficulty in performing its secondrole in the international financial system: defining a macroeconomic pro-

gram for a country in default. Argentina’s fall 2003 agreement with theIMF did not define the amount of primary fiscal adjustment Argentinaneeds to do beyond 2004 to repay all its creditors. Rather, the program leftthe job of negotiating Argentina’s fiscal adjustment path to Argentina’sprivate creditors. The funds Argentina owes the IMF have hamstrung theIMF and no doubt diminished its leverage. The IMF’s major shareholderswho—at least in 2003—did not always back the IMF in its negotiationswith Argentina have also weakened the IMF’s leverage. Eliminating theIMF’s preferred status would not solve these problems. Rather the solu-

tion is to find ways for the IMF to use more effectively the leverage thatcomes from its ability to lend when others will not.

The burden that priority payments to the IFIs places on a distresseddebtor should not be exaggerated. The IMF and MDBs typically are repaidin full only because they are patient enough to allow a country manyyears to recover before demanding repayment. They may even providenew money to cover interest payments in some cases.8 This is another waythe IFIs do not act like private lenders. Private lenders lend for short termsso that they have the option of getting out if conditions turn sour. TheIMF’s short-term lending is repaid only if the borrowing country is in de-cent economic and financial shape. Of course, defensive lending to reducethe risk of the debtor defaulting on the IMF—and also on the MDBs, be-cause an IMF program is a prerequisite for MDB lending—is hardly ideal.It ties up the IMF’s balance sheet and can lead to watered-down policyconditionality, reducing both the IMF’s credibility and its ability to lend to


8. IMF resources are at risk of default in the present system. A few countries, such as Sudan,

have defaulted on their IMF lending, though this happens rarely. The heavily indebted poorcountries (HIPC) process allows very poor countries to reduce their debt to multilaterallenders (IMF and World Bank) as well as their bilateral debt. Middle-income countries areexpected to pay the IMF and World Bank back in full over time, even if they default on theirprivate debts—and, in general, they do. Russia is a case in point.



other countries. The IMF is effectively “bailed in” if its short-term catalyticlending fails, as its short-term loans are rolled over to avoid formal ar-rears. Rather than getting absolute priority, the IMF and other IFIs oftenhave to struggle to convince the crisis country to continue to pay intereston the country’s multilateral debt.

Paris Club Debt and Private External Debt

The major bilateral creditor countries—OECD countries like the UnitedStates, Japan, the United Kingdom, and France—meet regularly in Paristo coordinate the restructuring of their debts to debtor countries. These bi-lateral creditors are commonly called Paris Club creditors.9 A debtor seek-ing to restructure the debt it owes Paris Club creditors is asked to seeka “comparable” restructuring from its other external creditors—be they bilateral creditors that are not part of the Paris Club or private lenders.Not surprisingly, the Paris Club has been a lightning rod for complaintsabout the official sector’s bail-in policies—in part because Paris Clubcreditors had significant exposure in the initial bond restructuring casesof Pakistan, Russia, and Ecuador.10

Private creditors accuse the Paris Club of being secretive, arbitrary, un-fair, politically biased, and unwilling to even talk to private investors (IIF1999b, Caplen 2000, and Booth 2001). These complaints can be separatedinto two critiques. One critique argues that the Paris Club should not aska debtor to seek a comparable restructuring of the debtor’s outstandinginternational bonds in order to facilitate the development of private capi-tal markets. A second critique calls on the Paris Club to do a better job of coordinating its restructuring with private debt restructuring. The firstcritique is wrong. The second has a grain of truth but is also based on anumber of misconceptions.

Why should the Paris Club condition its restructuring on a comparablerestructuring by other external creditors, including private creditors? Theanswer is simple: so as not to subordinate its taxpayers’ claims to those of other creditors and, in the process, subsidize payments to bondholders orother groups of creditors. The bilateral creditors who meet in the ParisClub do not lend to other sovereigns on the expectation that their debtswould be systematically subordinated to the sovereign’s other externaldebt, apart from debts owed to the IFIs. When most private lending toemerging-market sovereigns came from banks, the Paris Club condi-tioned its restructuring on the debtor’s willingness to seek a comparable


9. See Reiffel (2003) for a detailed history and analysis of the Paris Club and its procedures.

10. The Paris Club also had significant exposure in Indonesia, but bondholders did not. In-donesia’s creditors—largely Japanese banks—did not protest the Paris Club’s requirementthat Indonesia seek a comparable rescheduling of their claims. Indonesia’s bonded debt was

judged de minimis, and the Paris Club did not insist that it be included in the restructuring.



restructuring of its external bank loans. The banks met in a club of theirown—the London Club (which often meets in New York, despite itsname)—to coordinate their own restructuring. Bonds were typically leftout of these restructurings because the country’s bonded debt was toosmall to be worth the trouble (bonds were considered de minimis), not be-

cause the Paris Club agreed to give these claims priority. As international banks got out of the business of providing medium- and long-term sov-ereign financing and as sovereigns turned to the bond market to raisemoney, continuation of long-standing Paris Club policy implied that theParis Club would ask countries to seek to restructure their bonded debton comparable terms. The markets could and should have expected it.11

The argument that comparability is illegitimate because political ratherthan strict commercial considerations motivate official bilateral lendinglacks merit. Some bilateral lending certainly has been motivated as much

by a desire to aid a country as to be paid back with a profit. Other bilateralloans really are not even foreign aid so much as export subsidies. But evenin private-sector lending, the obligation to repay is independent of thereasons the lender made the loan. A financial conglomerate that makes aloan to win investment-banking business still has a valid claim. The in-terest rate the US government charges on its bilateral lending is intendedto reflect the risk of the taxpayer not being repaid, not the costs of sys-tematically subsidizing bondholders.12

The fact that Paris Club creditors have legitimate reasons for insisting

that a sovereign in distress treat all unsecured external debt comparablydoes not mean that they always have the leverage to assure that all suchdebt will in fact be treated comparably. Indeed, sovereign debtors oftenstop paying official bilateral creditors well before they stop paying privatecreditors, effectively treating Paris Club debt as junior debt. Strategic non-payment on Paris Club debts often helps a sovereign pay private creditorsin full and on time. For example, Nigeria accumulated over $23 billion of arrears to the Paris Club in the 1990s while paying its Brady bonds andother bonds in full.13 Ecuador was able to issue a new eurobond in 1997,

when it was in arrears to the Paris Club. Sovereign debtors know thatgoing into arrears to bilateral creditors has few immediate consequences.As a rule, bilateral lenders do not litigate to recover payment. However,


11. Some in the markets thought that the perceived difficulty of restructuring bonds wouldlead the Paris Club to exempt bonds from Paris Club conditionality.

12. After the credit reform, the charge is set in the Interagency Country Risk Assessment(ICRAS) process.

13. Former Nigerian President Sani Abacha and his clan are reported to have $2 billion intheir private accounts (“Sani Disposition,” The Economist, September 7, 2000); he and otherNigerian insiders are rumored to hold a substantial sum of Nigeria’s external sovereign debtin his private accounts. Nigeria’s long-standing desire to pay its private debt while not pay-ing bilateral debt was not simply the product of its belief in the sanctity of private contracts.



most countries eventually find that they cannot finance themselves indef-initely by running arrears to the Paris Club. When a country approachesthe IMF for financing, the Paris Club creditors—who also have the mostvotes on the IMF Executive Board—can insist that any IMF lending takeplace only in the context of a general restructuring of all the country’s ex-

ternal debt.Implementing the Paris Club’s policy of linking its own restructuring to

a comparable restructuring by other external creditors raises a host of dif-ficult practical questions. How can two sets of creditors reach agreementwith the debtor on terms that treat both sets “comparably” while also rec-ognizing their distinct preferences? This question does not just arise be-tween the Paris Club and private creditors or even just in sovereign re-structurings. Different groups of creditors in a corporate restructuringmay also have different preferences and want different restructuring

terms. In a wide range of restructurings, banks and retail investors put ahigher premium on preserving the face value of their claims than profes-sional money managers, who care more about the bond’s market valuethan its face value.

Private creditors’ complaints that the Paris Club initially did not makegreat efforts to explain how it went about assessing comparability havemerit, but a number of misconceptions have complicated the debate onthe relative treatment of Paris Club debt and private bonds. The first mis-conception is that different restructuring terms imply the absence of com-

parability. In some cases, private creditors have agreed to reduce the facevalue of their claims while Paris Club creditors have not. However, differ-ent terms can reflect the need to craft restructuring terms to match thepreferences of different creditors—not discriminatory treatment. For ex-ample, private creditors considered par Brady bonds (no face value debtreduction, low coupon) “comparable” to discount Brady bonds (face valuedebt reduction, higher coupon) in Brady restructurings.14

A need to minimize the budgetary costs that Paris Club creditors incurin a debt restructuring heavily (almost certainly too heavily) drives the


14. Private-sector representatives often ask for reverse comparability, with the Paris Clubmatching private-sector debt reduction if the private sector restructures before the ParisClub. Private creditors, of course, are free to hold up their own agreement on the sovereignuntil the sovereign is able to reach agreement with the Paris Club on terms that the ParisClub accepts and that private creditors judge comparable. Private creditors, however, have

been unwilling to delay a deal with mark-to-market gains to increase their leverage vis-à-visthe Paris Club. In any case such tactics would be unproductive. The Paris Club has differentpreferences and constraints than private creditors, and it should not agree to debt reductionsimply because private creditors are willing to do it. As a matter of policy, Paris Club credi-

tors should not agree to restructuring terms unless they are confident of getting the neces-sary budget funds to pay for the restructuring and budget funds are too scarce to be allo-cated solely on the basis of the amount of debt relief private creditors provide. Rather budgetfunds should be allocated on the basis of broader public policy objectives (including help-ing the world’s poorest).



club’s preferences. The accounting and budget rules for Paris Club debtrestructuring usually do not require that its creditors discount future cashflows at market rates. Paris Club creditors often can avoid taking a bud-getary charge even when they reschedule claims at relatively low interestrates for very long terms, because they can use a discount rate on the new

claims that does not reflect an objective assessment of the expected prob-ability of repayment. Consequently, Paris Club creditors have an institu-tional preference for the equivalent of the “par” option. Apart from a fewexceptions, the Paris Club generally does not provide debt reduction formiddle-income countries.15 However, few private creditors would preferthe terms of a typical Paris Club debt restructuring to those of the privatedeal, even when the Paris Club does not agree to debt reduction. Whenvalued at market discount rates, Paris Club restructuring terms usuallyimply significant reduction in the NPV of the creditors’ claims.16

Moreover, Paris Club loans include a series of features not present inprivate debt contracts. Paris Club debts effectively have an embedded op-tion that allows the debtor to roll over principal and capitalize interest atits own discretion—since the debtor can stop paying these debts with lit-tle consequence. Paris Club creditors do not litigate and never panic. Aprivate debt contract with no legal enforcement rights and similar embedded options would trade at a deep discount. These features imply thata Paris Club restructuring can be comparable to a private debt restructur-ing even when the terms differ—no private creditor would be willing to

provide credit to a sovereign at the terms, risk features, and spreads thatofficial bilateral creditors provide.

The second misconception is that private creditors in Russia andEcuador restructured their debts on substantially more generous termsthan the Paris Club. The relative treatment of Paris Club and private debtscan be assessed in different ways: Private creditors, not surprisingly, havetended to emphasize those ways of looking at comparability most favor-able to their arguments. In Russia, holders of Soviet-era external debt (re-structured in the London Club) did agree to do more than the Paris Club.

However, comparability needs to be assessed by looking at how Russiatreated all of its external private debts, including its Russian-era eurobonds. Since the high-coupon eurobonds were altogether excluded fromthe restructuring, the overall treatment of Russia’s private external debtwas comparable to that of Russia’s Paris Club debt. In Ecuador, a fair as-sessment of comparability requires looking at all the terms of the restruc-turing—cash sweeteners, maturity, coupon, and the treatment of collat-


15. Poland, Egypt, and former Yugoslavia are the most notable examples.

16. Paris Club creditors’ preferences often reflect the projected budgetary cost of the re-structuring, which differ from calculations of the NPV using market rates. Paris Club agree-ments are commitments to restructure on the proposed terms, but the actual execution of therestructuring depends on each country’s budget rules.



eral—not just the face value of Ecuador’s new bonds. Once adjustmentsare made for the debt reduction that stemmed solely from the early releaseof collateral on two of Ecuador’s Brady bonds, it is not obvious that un-collateralized private creditors offered more debt relief than Paris Clubcreditors. For example, the holders of Ecuador’s 2002 and 2004 eurobonds

were able to trade their existing claims at par for new high-coupon eu-robonds maturing in 2012.17 Private creditors tend to overlook cases likePakistan, where the terms of the private restructuring were more favor-able than those of the Paris Club restructuring. The Paris Club has beenfirm in insisting on the principle of comparability but flexible in interpret-ing its precise meaning.

Nonetheless, real differences—some of which are quite technical—make it difficult to coordinate the Paris Club and private restructurings:

The Paris Club typically restructures only those payments coming dueduring the country’s IMF program (in Paris Club terminology, the“consolidation period”). This can result in a series of restructurings of Paris Club claims rather than a comprehensive restructuring of the en-tire stock of Paris Club debt, which tries to address the country’s prob-lems once and for all. The Paris Club agrees to a comprehensive re-structuring of a country’s entire debt stock only after the country hascompleted a three-year IMF program. Private-sector debt restructur-ings, in contrast, typically restructure the entire “stock” in a single re-

structuring to avoid repeat restructurings. While bank restructuringsin the 1980s often were implicitly conditioned on the completion of anIMF program, bondholders generally have made an IMF program acondition for the execution of their restructuring.

There are no well-understood rules on how to allocate available near-term cash flow among different creditor groups. The Paris Club’s flowrescheduling model requires the country to allocate cash flows amongcreditors in proportion to the amounts they have coming due duringthe IMF program period. Paris Club creditors use this method to ap-portion payments among themselves. They turn to this method mostnaturally for apportioning payments between Paris Club and privatecreditors. This method tends to favor creditors with large arrears(often Paris Club creditors, because countries typically go into arrearson Paris Club debt before private debt) or with principal paymentscoming due during the consolidation period. Private creditors do not


17. Holders of uncollateralized Brady bonds did not get as good a deal as holders of the2002 and 2004 eurobonds, and holders of the uncollateralized portion of partially collateral-ized Brady bonds did agree to substantially alter the profile of their uncollateralized pay-ments to Ecuador’s benefit. However, all arrears on bonded debt were also settled in cash,while arrears to the Paris Club were settled in large part with new debt.



necessarily follow a similar rule. They often base restructuring termson aggregate amounts outstanding, not the amount coming due.

Paris Club processes are built around an assumption that the ParisClub restructuring precedes the restructuring of the debts of bothother bilateral and private creditors. The Paris Club uses its rules todecide what share of the sovereign’s near-term cash flow it shouldtake, leaving the remainder for other creditors to divvy up amongthemselves using their own rules. This process breaks down when pri-vate bondholders restructure before the Paris Club, as has been thecase in some recent restructurings.18

The Paris Club has its own arcane system for according informalseniority within the Paris Club in order to facilitate new lending. Acountry’s initial restructuring sets a cutoff date, and in theory lending

after the cutoff date is senior to old lending in the event of future pay-ments difficulties. This can become a problem if a country ends upgoing back to the Paris Club, because the original cutoff date does notchange. Private creditors do not have a comparable system or anycomparable complications. (In the 1980s, “new money” that bank cred-itors provided was initially kept out of the restructuring, but over timethis practice disappeared.)

There is value in improving dialogue between the Paris Club and pri-

vate creditors, so that both groups better understand how the other oper-ates and the sources of their different preferences. Indeed, the Paris Clubhas been taking steps to increase its transparency, after private creditorscorrectly pointed out that it was difficult to find information on how theParis Club operates or on the amount of debt a country owed to ParisClub creditors.19 However, the need to improve coordination does notmean that the Paris Club creditors should accept private demands to ne-


18. The difficulties in allocating near-term cash flows are a manifestation of the general dif-

ficulties created when one major group of creditors does a “flow” restructuring—i.e., it onlyrestructures those claims coming due in a defined period (Paris Club creditors do not accel-erate their claims)—and another group of creditors does a “stock” restructuring—i.e., it re-structures the entire stock of outstanding debt. The prorated distribution of cash flows ac-cording to claims coming due in the consolidation period works reasonably well if allcreditors are doing flow restructurings but less well if one creditor group is restructuring theentire stock. In practice, the Paris Club has been willing to interpret this rule flexibly.

19. The Paris Club set up a Web site, www.clubdeparis.org, that provides information aboutits activities, rules, upcoming cases, and claims being restructured. The Paris Club (oranother international body) could do more—for example, a more systematic registry of allParis Club bilateral claims might be useful to investors. Certain Paris Club procedures arearcane, and the flexibility that is consistent with Paris Club procedure—changing cutoff dates, for example—is not always obvious. More public information about its proceduresand rules might help everyone avoid misunderstandings.



gotiate restructuring terms on a case-by-case basis with a committee of private creditors.

The inherent difficulty in negotiations between governments and pri-vate financial firms means that there is a high risk of the “negotiations” re-sulting in a slower and more contentious process without necessarily pro-

ducing substantially different outcomes.20 Unless private creditors wantto lobby their legislatures for budget funding to pay for debt concessions,general Paris Club policies will dictate the club’s restructuring terms. Mostrestructurings follow the basic Paris Club procedures: normal terms (i.e.,substantial maturity extension at favorable interest rates but no debtreduction) for middle-income countries, Houston terms for poorer coun-tries, and HIPC terms for those that qualify for it. It takes a political deci-sion to provide a high-profile country with special treatment.21 The ParisClub is not a formal organization so much as an ad hoc group of rotating

creditors, with rules and procedures to enable relatively low-level techni-cal experts to handle a large number of cases reasonably efficiently. It isnot currently set up for high-profile negotiations with private creditors.22

This is not to say that the Paris Club’s specific practices do not needto change. For example, it is worth discussing how to better balance“flow” and “stock” considerations in allocating available near-term cashflows among various creditor groups.23 There is a case for favoring cred-itors who have been discriminated against before the restructuring andthus favoring those with large arrears (typically the Paris Club). There is

also a case for providing more cash to creditors who place the highestvalue on cash (market creditors) and who are most willing to agree to realdebt reduction in return for higher upfront cash payments.

The standard market critique that the Paris Club is a political wild cardwhose actions cannot be predicted, ironically, is not the strongest critiqueof the club. The Paris Club’s biggest problem is that its restructuring terms


20. Once the IMF defines the amount available for debt service and rules are agreed uponon the distribution of upfront cash, restructuring terms cannot be in sharp contrast to com-

parability and medium-term sustainability.

21. Special considerations were made for countries such as Poland, Egypt, and Serbia. Keycreditors have already committed to give Iraq special treatment.

22. France hosts the club and by convention both chairs its meetings and provides its secre-tariat, but the French secretariat cannot speak for all Paris Club creditors. Private creditorswould need to negotiate with the individual governments that make up the Paris Club, bar-ring a major reorganization.

23. We believe a widely understood rule—like some variant of the proportionality principleused in recent cases—would ultimately provide a more predictable and fair distribution of

the burden than a formal negotiating process, given the Paris Club’s need to operate with aframework defined by established rules and principles. All groups of creditors would needto reach an informal understanding that both Paris Club and private claims are rescheduledand/or paid in cash in proportion to the outstanding claims of both groups, either the over-all amount outstanding or the portion of claims coming due in a defined period.



are too predictable and fail to take the specific circumstances of the coun-try sufficiently into account. Paris Club rules reserve outright debt reduc-tion—which requires scarce budget resources—for the poorest countries.Middle-income countries get substantial cash flow relief but no debt re-duction. The absence of deep debt reduction from holders of either Ecua-

dor or Russia’s external bonded debt undermined private-sector com-plaints about the absence of Paris Club debt reduction in either country.24

Argentina, however, is seeking deep debt reduction from its private cred-itors. Since Paris Club exposure in Argentina is small, the absence of out-right reduction by Paris Club creditors won’t undermine the country’soverall solvency, but it still highlights how general rules can result in sub-optimal outcomes in specific cases. Fortunately, the G-7 has recognizedthat the Paris Club may need to be more flexible in some middle-incomecountries (Serbia and Iraq are obvious examples) and has called on it to

rely less on preset terms and more on debt sustainability analysis to fig-ure out how much relief a country truly needs (Group of Seven 2003b). Itremains to be seen how the Paris Club will use this new flexibility—andwhether private creditors will push the Paris Club toward restructuringterms that help them more than the crisis country.

Domestic Versus External Debt

The treatment of domestic debt is often the most contentious issue ina sovereign debt crisis. Sovereigns facing a deep crisis, like Argentinanow or Ecuador in 1999, rarely leave either domestic or external claimsentirely out of a restructuring. Consequently, the substantive policy issueis whether these two categories of debt should be restructured on similarterms or domestic debt should be given some priority.

What Is “Domestic” Debt?

The absence of a simple definition of the difference between domestic andexternal debt is one sign of the complexity of this issue. The standard eco-nomic definition is based on the principle of residency. External debt isany debt held by a foreign resident, regardless of debt instrument’s gov-erning law or currency of denomination. A local-law, ruble-denominatedRussian treasury bill held by foreign investors, consequently, counts as ex-ternal debt. Conversely, a dollar-denominated, New York–law eurobondheld by a Turkish bank should be considered as domestic debt. Anothereconomic definition uses the currency of denomination to define the dif-ference between domestic and external debt.


24. External investors who bought Russia’s domestic GKOs clearly did take substantiallosses.



The legal definition, in contrast, focuses on governing law. Domesticdebt is defined as debt governed by domestic law, regardless of whetherthe debt is denominated in local or foreign currency or whether a foreignor domestic resident holds it. Conversely, debt governed by foreign law isdefined as foreign debt, even if a domestic resident holds it. Phrases like

“domestic banks hold lots of the country’s external debt” implicitly relyon governing law to define what constitutes external debt (IMF August2002).

In most countries, domestic residents still hold a larger fraction of thedomestic-law debt than foreigners, and nonresidents still hold a largerfraction of the foreign-law debt. As markets become more integrated,however, foreigners are increasingly investing in domestic-law debt anddomestic residents are increasingly buying the country’s foreign-lawdebt. The banking systems of Turkey and Lebanon hold a substantial frac-

tion of these countries’ New York-law eurobonds—as did Argentina’s banking system before the November 2001 eurobonds-for-guaranteed-domestic-loans swap. In other cases—the United States today and Russiain 1998—foreign investors hold large quantities of a country’s local-law,local-currency debt.25

The difference between the legal and economic definitions of domesticand external debt can complicate the discussion of the relative treatmentof domestic and external creditors. If foreign investors hold the country’sexternal-law debt, and domestic investors hold the domestic-law debt,

it is technically easy to offer different restructuring terms to holders of domestic and external debt. Payments on domestic-law debt are usuallymade inside the crisis country and are harder for holders of foreign-lawdebt to stop. If both domestic and foreign residents hold the same external-debt instrument, offering different terms to residents and nonresidents ismore difficult.26 It typically requires a multistep process. Argentina’s No-vember 2001 “bonds for loans” swap, for example, was designed to ap-peal to domestic holders of the country’s eurobonds (foreign investorstypically could not hold a domestic-law loan). Once most domestic resi-

dents exchanged their foreign-law bonds for a new domestic instrument,it was possible to treat the new domestic instruments differently than theoriginal bonds.


25. In practice, it is difficult to track who holds a sovereign’s external-law bonds. Manycountries simplify the calculation of their country’s external debt by just assuming that allexternal-law bonds are held externally. As a result, the reported external debt of some coun-tries exceeds the amount that nonresident investors truly hold. This was particularly truewith Argentina.

26. The sovereign often does have the ability to offer different terms to external and do-mestic investors if they both hold the same domestic debt instrument—e.g., Russia offeredinternational investors holding its GKOs different terms than domestic investors. The legalleverage of external investors in domestic debt markets varies dramatically from country tocountry.



debt restructuring that leads to a banking crisis typically reduces the valuethat external creditors can expect to receive.

The first argument highlights the need to distinguish between the adjustment effort foreign investors and domestic agents make, not between therelative treatment of foreign and domestic creditors. The domestic resi-

dents in a crisis country would bear a substantial burden even in the un-likely event that their financial claims on the government were not re-structured. Domestic fiscal adjustment requires an increase in taxes andcuts in public spending and services. Higher direct taxes (such as incometaxes) and indirect taxes (such as consumption/sales taxes) and reduc-tions in public salaries, public pensions, and public services all result inlower income for domestic residents. Also, residents mostly bear the fallsin output, employment, and consumption needed to improve the coun-try’s external balance. Falls in the local-currency value of a wide range of

real and financial assets, such as equity and real estate, also tend to havea bigger impact on residents than nonresidents, because residents typi-cally hold more such assets than foreign investors.27

In most cases, however, the financial claims of domestic residents alsoare restructured. A government that can no longer act as a lender of lastresort to the banking system often has to freeze deposits, turn short-termdeposits into long-term bonds, implement capital controls and otherwiserestrict access to domestic financial assets. Frozen deposits and otherdomestic debts are often restructured in ways that imply a fall in their

real value. Indeed, reductions in the real value of domestic debts can bethought of as just another way of taxing residents. The tax—formallycalled a capital levy—is imposed on the value of financial assets ratherthan on current income or spending.

Assessing whether domestic residents and foreign investors are makinga “comparable” contribution to crisis resolution therefore requires looking beyond the relative treatment of domestic and external debt and examin-ing the many other ways in which domestic residents contribute to crisisresolution. Of course, there is no single, unitary domestic agent that bears

all domestic costs. The distribution of the domestic adjustment burdenamong domestic taxpayers, domestic beneficiaries of public spending, anddomestic holders of financial assets is usually at least as contentious as therelative treatment of external and domestic debt. Conceptually, though, itis still possible to separate the adjustment costs domestic agents bear fromthe concessions foreign investors make.

A truly insolvent country cannot pay its foreign debt in spite of makingthe maximum feasible domestic adjustment. Foreign investors, therefore,have to bear the full residual burden to restore solvency. The improvement


27. The real value of other local currency–denominated financial assets (like those a debtcontract creates between two citizens) can also fall, though a fall in the real value of privatecontracts can benefit debtors even as it hurts creditors.



in the country’s fiscal position that can be generated by reducing the valueof domestic public debt—effectively a tax on financial wealth—should bethought of as part of the maximum feasible domestic adjustment. Just asother types of domestic adjustment have limits, so does the amount of adjustment that can come from a haircut on the domestic debt. From this

point of view, the ideal debt restructuring process would start by deter-mining how much domestic pain/adjustment is economically necessary—as well as how much is politically/socially feasible—to put the country back on a sustainable growth and fiscal path. This in turn would deter-mine the debt relief needed from external creditors. The discussion aboutdomestic debt restructuring should be part of the internal debate over whowithin the society should bear the needed domestic adjustment cost, notpart of the discussion over external debt restructuring terms.

It is worth noting that domestic debt reduction has only a limited impact

on the economy’s overall capacity to generate the foreign exchange forexternal debt service. Payment of the government’s external debt requires(1) that the economy as a whole generate the needed foreign currency and(2) that the government raise enough money domestically to buy this for-eign currency. Ahaircut on domestic debt would reduce domestic financialwealth, leading to lower consumption and imports. However, this is a veryindirect way of generating the needed external adjustment. Of course,smaller domestic debt payments mean that the debtor can use more taxrevenue to purchase the foreign exchange needed to make external debt

payments. But large domestic haircuts may lower the amount of domesticfiscal adjustment that is politically sustainable even as those haircuts re-duce domestic financial claims on that fiscal surplus. Domestic residentswho have taken large financial haircuts on their domestic debts may not bewilling to accept additional sacrifices (higher taxes/lower government ser-vices) to pay foreign creditors (see box 7.1 for an elaboration).

Difficulties That Domestic Banks Create

The argument that a domestic debt restructuring is a form of domestictaxation that redistributes domestic resources within the economy andtherefore should be considered a form of domestic adjustment hinges onrather abstract questions of equity. The banking sector’s heavy domesticdebt holdings more commonly justify the different treatment of domesticand external debt.

Banks are almost always highly leveraged: A bank’s owners put up cap-ital, and the bank takes in short-term deposits and invests the resulting

funds in longer-term financial assets. The mismatch between a bank’spromise to return depositors their funds on demand and its longer-termand often illiquid assets adds to the financial fragility inherent in a highdegree of leverage. If the value of the financial assets the bank holds falls




a little, then bank capital can be written down and depositors can be keptwhole without any injection of taxpayer funds. If the fall in the value of the bank’s assets exceeds bank capital, in theory, depositors need to takea haircut. Yet they rarely do. Banks play a key role in the payments sys-tem, and an uncontrolled bank run triggered by fears of losses among de-

positors is economically devastating. Consequently, the government al-most always steps in to protect bank depositors from losses in the eventof a systemic crisis, even in the absence of formal deposit insurance.

Governments typically bail out banks by giving them—perhaps in ex-change for a portion of the banks’ equity—a government bond of sufficientvalue to balance the banks’ liabilities and assets. These long-term bonds area way of protecting deposits from losses by shifting the losses onto futuretaxpayers. If the domestic banking system holds a large amount of domes-tic debt and the government wants to protect depositors from losses, a hair-

cut on domestic debt implies the issuance of more domestic debt to bail outthe banks. Consequently, imposing an equal haircut on foreign bonds andthe domestic bonds held by local financial institutions only increases theamount of new domestic debt that the government will have to issue to fi-


Box 7.1 Different haircuts for domestic and external creditors

Consider the following illustrative and hypothetical example. A government has a debt-

to-GDP ratio of 100 percent, with domestic creditors holding one half and foreign cred-

itors the other half. This debt is unsustainable, and the country formally defaults. At the

time of default, the country was not running a primary surplus. All agree the government

needs to tighten its belt, but it cannot achieve a primary surplus of more than 4 percentof GDP. All also agree that the long-run growth rate of the economy will be 3 percent

and that the long-run real interest rate will be 11 percent. The differential between the

real interest rate and the growth rate is 8 percent (11 percent minus 3 percent), which—

in conjunction with a primary balance of 4 percent—implies that the debt-to-GDP ratio

should be reduced to 50 percent.

One solution would be to cut everyone’s debt in half, so that of the 50 percent of GDP

in new debt, one half is held domestically and the other half externally (scenario 1).The

domestic haircut is a form of taxation—a capital levy—that is imposed on domestic fi-

nancial assets. Suppose that domestic investors, who are also taxpayers, decide that

they would prefer to spread their losses over time (scenario 2).1 They consequently

agree to pay higher taxes to produce a larger surplus to avoid reducing the face valueof the debt they hold. If foreigners take the same haircut as before, the country’s debt-

to-GDP ratio falls from 100 to 75 percent. Domestic investors still hold debt worth 50 per-

cent of GDP, while foreigners now hold debt equal to 25 percent of GDP. The primary

balance required to stabilize the debt ratio is higher, since the debt level is higher—6

percent of GDP. Formally, scenarios 1 and 2 treat both domestic and external investors

the same, even though the second scenario does not impose a haircut on domestic

creditors. In both cases, foreigners hold debt equal to 25 percent of GDP after the re-

structuring, and 2 percent of GDP is allocated to service this external debt. Domestic in-

vestors are also agreeing to the same losses: A large domestic haircut implies that the

full burden of domestic adjustment is being born upfront in the form of a capital levy

(box 7.1 continues next page)



nance the bank bailout. Reducing the value of banks’ holdings of domesticgovernment bonds—so long as explicit or implicit government deposit in-surance fully backs the banks—provides almost no debt reduction for thegovernment.28 If depositors cannot take losses and bank capital is ex-hausted, then taxpayers have to bear the burden of bailing out the banks.

In theory, the effective haircut that foreign investors suffer is equal inthe following two scenarios:

Foreign investors are discriminated against in the debt restructuringand receive worse terms than domestic banks holding the same bonds;

Foreign investors are not discriminated against, and both domestic banks and foreign investors receive the same terms, but a much higheraverage haircut is required since the resulting bank bailout will re-


while a larger primary surplus implies that the same loss is spread over time.2 The size

of the primary balance devoted to service the foreign debt determines the size of the for-

eign haircut, regardless of how domestic losses are allocated over time.

Let us take the discussion a step further. Suppose that foreign creditors convince thecountry that a 50 percent haircut on their debt is too much, and the country agrees to

impose a proportional haircut of 25 percent on both foreign and domestic creditors.The

debt ratio then falls to 75 percent: Foreign creditors hold debt equal to 37.5 percent of

GDP, as do domestic creditors. A primary balance of 6 percent of GDP is now required

to sustain the higher debt burden that emerges from a smaller haircut (scenario 3). Rel-

ative to scenario 2, the country is running the same 6 percent primary balance, and ex-

ternal creditors are taking a smaller haircut. This is possible because domestic investors

are in effect agreeing to accept larger losses. Half of the 6 percent primary balance, or

3 percent of GDP, is devoted to servicing the foreign debt, relative to only 2 percent of

GDP in the previous two scenarios. In scenarios 2 and 3, the country is running a 6 per-

cent of GDP primary balance, but the 6 percent primary balance in the second scenario

was predicated on the need to spread the domestic losses in the first scenario over

time, not on the need to do more domestic adjustment to service external debt.

1. Here we assume that the average domestic creditor is equivalent to the average

domestic taxpayer. This is true on average, but individual differences imply only some

domestic redistribution.

2. On an NPV basis, a 50 percent haircut and a primary balance that increases from 0

to 4 percent leads to an equivalent loss to domestic residents as a primary balancegoes from 0 to 6 percent and no haircut.

Box 7.1 (continued)

28. The same argument holds if the haircut is imposed on the government debt holdings of pension funds whose benefits are defined. Unless the government wants and can impose acapital levy on the workers who are entitled to these assets, any haircut on these claims will

become another liability for the government.



quire the government to issue new debt to protect bank depositorsfrom losses.

The advantages appearing to treat all creditors the same and then re-capitalizing the banking system have to be balanced against the advan-

tages of avoiding the bankruptcy costs associated with a systemic bankrestructuring. Restructuring all debts avoids protecting both wealthy do-mestic holders of domestic debts and the banks’ equity owners from thepain of a restructuring. However, pushing the entire banking system intoinsolvency through a comprehensive restructuring is also costly. Restruc-turing all debt on equal terms increases the odds of a deposit run.29 It iseasier for the government to tell depositors not to worry if banks’ debts are being excluded than to explain that it will issue more debt to make up anylosses in the banking system from a debt restructuring.

One could certainly argue that domestic depositors should bear someof the burden in a sovereign debt crisis and that the government shouldimprove its own solvency by imposing a capital levy on the financialwealth of all bank depositors. While domestic debt is rarely restructuredon the same terms as external debt, domestic debt is usually restructuredin severe crises, and bank depositors often take losses. Deposit freezes,forced conversion of foreign-currency deposits into local-currency de-posits (pesification) and of deposits into longer-term bonds (at a marketvalue well below par), and caps on the interest rates provided to frozen

deposits are all ways of imposing a haircut on bank deposits.The choice between imposing a tax on bank depositors and asking ex-ternal creditors for additional debt relief illustrates well the basic argu-ments for giving more favorable treatment to domestic debts. Bank depos-itors are likely to see their real incomes fall no matter what as domesticoutput falls and may also bear most of the burden of the tax increases andspending cuts needed to improve the government’s fiscal position. Aslong as the amount of pain that domestic agents, including bank deposi-tors, are willing to take is limited, only the country’s nonresident creditors

can bear the residual adjustment burden. Of course, this creates problemswhen there is a large gap between the amount of “pain” domestic agents


29. From an intertemporal perspective, a country’s ability to service its defaulted externaldebt is independent of whether the depositors accept a capital levy or not. A haircut or cap-ital levy on deposits results in immediate losses for domestic residents. If there is no deposithaircut, then the banks will have to be recapitalized with recapitalization bonds that have to

be serviced over time through higher taxes/revenues. From an intertemporal point of view,the domestic burden of an immediate capital levy on deposits should be equal to the dis-counted value of the higher future taxes needed to pay for bank recapitalization (leaving

aside distribution effects). Consequently, external creditors should not necessarily care howthe country decides to resolve the insolvency of its banking system. If the average taxpayeris also the average depositor, this intertemporal shift of the burden has no domestic distri-

butional consequences. Realistically, some redistribution of wealth will occur in the two al-ternative scenarios, as individual depositors may differ from the average taxpayer.



are willing to bear—whether through more fiscal adjustment or a deeperdomestic debt restructuring—and that international investors are willingto accept. Unless the two sides reach agreement on how to apportion theeconomic loss, no deal is possible. Deals where creditors pretend that theywill be paid what they think they deserve at some point in the future and

where debtors pretend that these payments can be made without any ad-ditional adjustment carry a high risk of sowing the seeds for future trouble.

The final balance between domestic adjustment of all sorts and externaldebt relief will reflect the economic, political, and legal leverage of all par-ties in the process. Chapter 8 emphasizes that the legal leverage of externalcreditors is limited so long as the debtor is willing to incur the economiccosts (loss of external market access and difficulty attracting new domesticas well as external investment) of prolonged default. Domestic creditorsmay not have greater legal leverage but typically do have greater political

leverage: Any elected government is likely to prefer prolonged external de-fault to a prolonged domestic bank holiday. Domestic creditors also area far more likely source of new financing than external creditors. Finally,domestic creditors often—though not always—have less diversified port-folios than the average international investor.30 They consequently oftenhave a greater stake in the outcome of the restructuring.31 The equilibrium between domestic and external adjustment—and the relative priority at-tached to paying different sets of claims—will reflect the government’sperception of the relative costs and benefits of all these factors.

A Real World Example: Argentina

Argentina offers a concrete example of many of the issues alreadydiscussed.

blurring the lines between domestic and external debt before the crisis. Ar-gentina entered its crisis with a relatively large stock of external-law bonds and a relatively small stock of domestic-law debt. Domestic banks, pension funds, and individual Argentines were large holdersof Argentina’s external-law bonds. During its crisis, Argentina con-


30. An investor holding a diversified portfolio of emerging-market debt can absorb largelosses at one point in time in one emerging-market economy without necessarily takingoverall losses. Indeed, studies (such as a recent one by Klingen, Weder, and Zettelmeyer2004) suggest that over long horizons, emerging-market investors receive on average returnsequal to a risk-free rate asset such as US Treasury bonds. In 2001, returns on the emerging-market bond index (EMBI) were positive despite Argentina’s default.

31. There are cases, though, where some external investors have a less diversified portfoliothan some domestic investors. A retail Italian investor that put—unwisely—all its savingsinto Argentine bonds may have more “Argentine” exposure than a wealthy Argentine witha large offshore bank account, of which only a small proportion was invested in Argentina’sexternal debt.



verted much of its external-law debt into domestic-law debt througha “eurobonds for guaranteed loans” swap and added to its domestic-law debt by issuing new compensation bonds (Bodens) to make up forthe banks’ losses. Domestic investors, however, are estimated to holdat least 40 percent of Argentina’s remaining external-law bonds.

difficulties assessing equity: When should the clock start? During the latterpart of 2000 and throughout 2001, Argentina could not raise fundsabroad. However, it could still place significant quantities of newexternal-law bonds with its domestic banks and pension funds. Fundsraised domestically—along with IMF lending—helped to finance in-terest payments as well as the repayment of maturing principal on theportion of Argentina’s external-law debt that was held abroad. Do-mestic residents were effectively forced to increase their exposure tothe government of Argentina even as international bondholders were

reducing their exposure. After Argentina’s default, however, the do-mestic debt holders that accepted the restructuring terms implicit inthe “pesification” have been paid while the external-law bonds remainin default. External creditors can complain that Argentina has givenpreference to domestic debt after its default; domestic creditors canlegitimately respond that Argentina’s crisis started well before its ex-ternal default and that the “clock” should have started ticking whenfunds raised domestically were used to pay external creditors.

limiting the banking system’s losses after a sovereign default. Argentina’scrisis was so severe that it was impossible to protect bank depositorsfrom any losses. After Argentina’s default, its domestic-law debt, in-cluding the guaranteed loans that were created in a “bonds for domes-tic loans” swap only weeks before, was redenominated from dollarsinto pesos by government decree. Even though domestic bank depositswere converted into pesos at a relatively favorable 1 to 1.4 peso ex-change rate and then indexed for inflation, depositors clearly havetaken losses. Those who kept their claims in the banking system after

the deposit freeze was lifted now have a deposit worth about 60 centsrather than a dollar—and those who pulled their funds out at the heightof the crisis often took larger losses. Yet there is also little doubt that Ar-gentina has put a higher priority on resolving its domestic banking cri-sis than on resolving its external-debt crisis and has sought to limit thescale of losses domestic depositors have borne. The banks’ domesticdebts have been serviced according to the new restructuring termswhile other debt has remained in default. The government also has is-sued new debt (Bodens) to avoid the need for depositors to take larger

losses as a result of Argentina’s household and corporate debt crisis. the difficulty in solving a debt problem by default if domestic banks hold large

amounts of debt. Ironically, while the government of Argentina de-faulted in part because it had too much debt, its default has resulted




in an increase in its overall debt stock. Since Argentina’s debts have in-creased by more than its capacity to pay, it has divided its debt intotwo categories—one group that has been given priority and anotherthat has not. Argentina paid the IFIs, the banks’ pesified guaranteedloans (before being restructured twice, these claims were international

bonds) as well as the new Bodens thoroughout its crisis. It has been indefault on bonds held by external investors and by Argentina’s pen-sion funds since early 2002. (An exchange offer for these bonds is ex-pected in the fall of 2004.)

the difficulties in resolving a systemic banking crisis. Argentina had botha sovereign crisis and as a result of its private firms’ extensive dollar-denominated debts, a systemic corporate crisis. Argentina moved pre-emptively to solve its domestic corporate payments crisis by convert-ing domestic bank loans into pesos at a favorable 1 to 1 exchange rate.

The gap between the 1 to 1 conversion rate on banks’ loans and the 1to 1.4 conversion rate on deposits initially resulted in large paperlosses in the banking system, but the government subsequently pickedup most of the tab by giving the banks new bonds to compensate themfor their losses in this “asymmetric pesification.” Many have criticizedthe asymmetric pesification. However, those who blame pesificationor asymmetric pesification for the financial chaos of the crisis are con-fusing cause with effect. Almost all Argentine households and firmshad more dollar debt than they could pay following the devaluation.

Systemic corporate and banking crises are always costly, and the re-structuring process itself is almost always politicized. Large financiallosses need to be apportioned among important domestic constituen-cies: domestic depositors, domestic taxpayers, beneficiaries of govern-ment spending, domestic banks, and domestic businesses.32

The last point is worth exploring in more depth, because the cost of Argentina’s domestic banking and corporate crisis has dramatically in-creased Argentina’s debt stock. There are two distinct issues. First, could

a different way of resolving Argentina’s domestic crisis have reduced theoverall losses associated with its crisis? Second, could losses have been al-located in a different way, so that Argentina’s taxpayers—and perhaps itsexternal creditors—picked up a smaller tab?

In our view, the large losses of Argentina’s banking system—losses thatthe banks’ owners, bank depositors, or taxpayers had to absorb—are notprimarily the result of pesification or even the asymmetries of the pesifi-cation process. Any banking system that has lent to domestic borrowersin dollars—or another external currency—takes large losses in the event


32. In Argentina, external investors owned many local banks and local utilities with largedollar debts—a fact that would have complicated any process for resolving the country’sdomestic crisis.



of a major currency depreciation. Argentina had a particularly large prob-lem because its domestic dollar debts were unusually large and its exportsector was unusually small (Goldstein and Turner 2004). It is, of course,impossible to know for sure if the losses from pesification are larger,smaller, or equal to the losses that would have occurred if dollar claims

had remained in dollars and been restructured. Yet there is good reason to believe that keeping all bank loans in dollars would have been as costly,in aggregate.

Indeed, if all domestic loans had been kept in dollars and the peso/dollar exchange had still fallen from 1 to 3 (or even a bit less), almost allthe financial assets of the banking system—loans to the government,provinces, households, and firms—would have fallen into default.33 Thiswould have put immeasurable strain on the banking system, the domes-tic bankruptcy system, and many small firms and households. The result-

ing restructuring process would have been both lengthy and costly. Indi-vidual debt renegotiation is slow: Large domestic Argentine firms who borrowed from abroad and thus did not benefit from pesification gener-ally were still in default at the end of 2003—two years after the crisis.Moreover, it is unrealistic to believe that the government would not have been drawn into the process of renegotiating household mortgage andsmall business loans. Some across-the-board solution was clearly neededfor small borrowers, and deciding who should get how much relief—andwho should bear the associated costs—would have been politically para-

lyzing even if the debts had initially remained in dollars.Another option would have been to keep domestic loans in dollars but

to reduce all dollar debts by a fixed percentage. This option, however, isnot all that different from pesification. One to one pesification is similar toa two-thirds haircut of the face value of all domestic dollar debts. Ar-gentina could have kept debts in dollars and imposed a smaller haircut.However, there is no guarantee that this would have reduced overalllosses.34 In the face of a deep recession, less debt relief would have impliedsome additional firms and households could not service their debts.35 The


33. Pesification prevented domestic loans to the government and provinces from falling intodefault, though the new debt carries a low interest rate. Households, consumers, and small

businesses whose debts were pesified are generally paying on their peso debt. Argentina’s banking system still has problems—the yield on government debt is low, and probably below the average long-run cost Argentine banks will need to pay on their deposits. Many big Argentine firms also remain in default. They cannot resume payments on their domesticdebts before reaching agreement on restructuring terms with their external creditors.

34. Pesifying both domestic loans and deposits at a 1 to 1.4 rate would have reduced the sizeof the across-the-board haircut—and probably resulted in a higher ratio of nonperformingloans. Pesifying both domestic loans and deposits at a 1 to 1 rate would have provided bor-

rowers the same relief but imposed bigger losses on bank depositors.

35. Remember that the revenues of domestic firms without export revenues increased in linewith the growth of nominal domestic GDP, so pesification did not necessarily reduce the realvalue of their debts by two-thirds.



marginal cost of pushing additional households and firms into bank-ruptcy should not be underestimated: Bankruptcy itself is costly, and anunresolved debt overhang can destroy value. Some firms that paid theirpesified debts in full—and effectively paid their original dollar debts inpart—would not have paid anything at all for an extended period. Pro-

viding debt relief through one to one pesification produced winners andlosers (sometimes arbitrarily); it was done in a haphazard way and like allacross-the-board solutions, provided some debtors with more relief thanthey needed and others with less. However, the scale of Argentina’s do-mestic debt problem was such that it is not clear that any alternative solu-tion would have reduced the overall size of the financial loss—though un-doubtedly it could have produced a different set of winners and losers.36

The second question is the distribution of the resulting loss. It is hard tosee how more of the loss could be passed on to the Argentine banks’ own-

ers. The restructuring process effectively wiped out all of the banking sys-tem’s existing capital and inflicted large losses on the banks’ owners (oftenmajor international banks).37 The government has had to issue new bondsto the banks to offset the losses created by the asymmetries in the pesifi-cation process precisely because foreign bank owners otherwise wouldhave walked away, leaving the government in control of almost the entire banking system. If providing less debt relief to domestic debtors wouldhave led only to larger losses from nonperforming loans, as we have ar-gued, then reducing the costs that taxpayers picked up would have re-

quired shifting more of the costs on to domestic depositors. The “asym-metric” part of asymmetric pesification was a way of limiting the domestic


36. Indeed, the Argentine pesification was not substantially different from the US decisionto repudiate the gold clause when the country went off the gold peg in 1931, and the dollardepreciated against gold by two-thirds. The US Supreme Court upheld the repudiation of the gold clause in 1935. Without such repudiation, large segments of the corporate and busi-ness sectors whose debts were contractually linked to the value of gold would have gone

bankrupt. Argentina “dedollarized” its domestic debts, and the United States “degoldized”its domestic debts. See Kroszner (1998) for more details.

37. The overall impact of the steps Argentina has taken to restructure its banking system onthe interest of the banks’ foreign owners is complicated, since some steps helped the banks’owners and hurt others. While the asymmetries in pesification initially imposed large losseson the banks’ owners, but the banks subsequently have been largely, though not yet fully,compensated for the effects of asymmetric pesification. Some banks are probably better off with pesified deposits, a performing pesified loan, and a compensation bond than a non-performing, dollar loan, and dollar deposits. Foreign-owned banks may have been hurtmore by the freeze on utility tariffs, which has damaged the financial health of many of theirmost important clients, than by asymmetric pesification plus compensation. Moreover, thegovernment’s overall approach to the resolution of the banking crisis undoubtedly has had

a large and often unexpected impact on the competitive position and balance sheets of dif-ferent banks. The government has not used the crisis to close, consolidate, intervene, or oth-erwise restructure a number of publicly owned banks that went into the crisis in poor fi-nancial condition, and bank regulators have not forced the closure of weak, domesticallyowned private banks.



depositors’ losses by having taxpayers pick up the difference between theconversion rate on bank deposits and bank loans. Asymmetric pesificationis not all that different from reducing all dollar deposits by a smaller per-centage than all bank loans, with taxpayers making up the difference.

Politically, though, it seems unlikely that it would have been possible to

increase the losses borne by domestic bank depositors through a differentdomestic bank restructuring process. Most bank depositors do not believethat they are the beneficiaries of Argentina’s system of assigning priority.Rather, they believe that they have taken unwarranted losses because the banks failed to honor their dollar deposits in full. This highlights our broader argument: If Argentina had kept the bank deposits in dollars (orconverted dollar deposits into peso deposits fully indexed to the dollar),depositors would have borne fewer upfront losses while the governmentof Argentina would have had an even larger bank bailout bill. Rather than

taking losses upfront, Argentines would have spread the banking sys-tem’s losses over time, through higher taxes and less government spend-ing to pay for the bank bailout. On the other hand, had the banks not beenrecapitalized through the compensation process and had depositors takena larger haircut, depositors would have less financial wealth, the govern-ment of Argentina would have a smaller domestic debt stock, and Ar-gentina would have less need to either reduce government services or in-crease taxes to pay its debts.

External creditors argue that they—and not future Argentine taxpay-

ers—are currently being asked to pay some of the cost of the bank bailout:Argentina is asking external creditors to agree to a large haircut to makeArgentina’s debt stock compatible with its current fiscal effort. This argu-ment has a grain of truth but also makes a number of assumptions thatshould be laid out explicitly. A larger deposit haircut helps external cred-itors only if there is no offsetting reduction in the size of the primary sur-plus that Argentines are willing to run. Less domestic debt and the sameprimary adjustment imply that more of the primary adjustment would beavailable to provide the fiscal resources (though not the foreign exchange)

needed to support a higher level of external debt service. However, it isentirely possible that a bigger domestic haircut would result in a reduc-tion in Argentina’s political willingness to adjust, a smaller primary sur-plus, and no more resources for external debt service.

Concluding Remarks on Domestic Debt

A sensible system of priorities should recognize that domestic and exter-

nal debt have a number of important economic differences and that thismay call for different—and sometimes better—treatment of certain do-mestic financial claims. Domestic debt restructuring is effectively a tax onone kind of domestic financial wealth and should be considered part of




the internal adjustment the debtor needs to make to restore itself to sol-vency, not part of the external debt restructuring. However, our argumentfor treating domestic debt differently from external debt is not an argu-ment that the debtor should avoid making any domestic adjustments. Theoverall domestic adjustment effort that the bankrupt country makes and

the concessions external creditors provide should be balanced. A bank-rupt sovereign that pushed all the costs on its external creditors to avoidany domestic adjustment would obviously fail to provide the needed bal-ance. The case for treating domestic debt better is that domestic residentsare adjusting in other ways, not that external creditors should spare a cri-sis country’s residents from any painful domestic adjustment.

The economic case for treating domestic debt differently than externaldebt is particularly persuasive if the financial system holds a large shareof the domestic debt. The economic disruption associated with the total

collapse of the domestic financial system is well worth avoiding. How-ever, there are also limits to how much the sovereign can favor the do-mestic banking system. Domestic debts often need to be rescheduled toaddress liquidity problems, and if the domestic debt stock is large, thendomestic holders of government debt also will likely need to accept areduction in their financial wealth as part of the overall domestic adjust-ment required to restore sustainability. If domestic banks hold all the gov-ernment’s debt, then it is obvious that domestic debt the banks hold can-not be fully protected from losses.

Arguments in Favor of a Formal Debt Seniority Regime

The uncertainties about the relative priority accorded to different claimson a sovereign in the current sovereign debt restructuring process—un-certainties that are obvious in Argentina—have prompted a number of calls to develop a clearer system of priorities for sovereign claims.38 Pro-ponents of reform argue that a set of more defined priorities would im-

prove on the current nonregime in several ways:

First, the ambiguity in the current system makes it much harder torapidly reach agreement on a needed debt restructuring. The systemof priorities in a bankruptcy regime makes both the debt restructuringmore orderly ex post and the relative treatment of different claimsmore predictable ex ante (Gelpern, 2004). In the absence of an agreed,enforceable system of priorities, different groups of creditors and thedebtor have to agree (whether explicitly or by default) on the relative

priority that should be given to different sets of claims before they cannegotiate actual financial terms.


38. See Zettelmeyer (2003) and Gelpern (2004) for thoughtful discussions of seniority issues.



Second, the current system risks encouraging hard-to-restructure debtcontracts. Creditors can game the system by hoping the debtor willgive informal priority to instruments that are legally hard to restruc-ture.39 Providing formal seniority to the various sovereign debt claimsmight allow the development of instruments that make debt restruc-

turing less messy (Zettelmeyer 2003). Third, the lack of a formal seniority structure in sovereign debt can

distort the sovereign’s incentive to manage its debt wisely (Bolton2003, Bolton and Skeel 2003, Paasche and Zin 2001). The absence of afirst to lend, first to be paid—or first in time—priority structure canlead to sovereign overborrowing. The first creditor who lends to a sov-ereign debtor would prefer that the sovereign not take on any moredebt, thus assuring that the initial creditor would have an exclusiveclaim on the sovereign’s debt-servicing capability. Additional loansfrom other creditors dilute the value of the first creditor’s claim unlessevery new loan results in an equal increase in the debtor’s overall abil-ity to pay. The first lender will anticipate this risk and therefore chargethe sovereign more, while later creditors will charge less than theyshould because they can appropriate some of the sovereign’s debt-ser-vicing capacity from earlier creditors. The lender’s inability to controllater borrowing imposes an externality on earlier borrowing. If earlierclaims had greater seniority than later claims, the spread on the initialdebt would be lower and that on later debt would be higher. Thiswould reduce a sovereign’s incentives to overborrow.

Obstacles to the Creation of a More Defined System of Priorities

Difficulty in agreeing on the relative priority that should be given to dif-ferent claims on the sovereign is certainly a major impediment to rapidagreement on a restructuring. Yet the conceptual and practical obstacles tocreating an explicit seniority system for sovereign debt are profound.

Diversity of Sovereign Debt

The main obstacle is the sheer breadth and complexity of the financialclaims on the sovereign and more broadly, the number of “stakeholders”that have an economic interest in how the sovereign’s difficulties are re-solved. Ranking the seniority of all the explicit claims on the sovereign(domestic and external debt and within external debt, the claims of dif-


39. The best example of this may be Nigeria’s Brady bonds, which have several features thatwould make Nigeria usually vulnerable to the risk of litigation, should it default on itsBrady bonds. On the other hand, Nigeria traditionally has had other reasons to pay its Brady

bonds while running arrears to the Paris Club. Wealthy Nigerians, including former Presi-dent Abacha’s family, reportedly had large holdings of Nigeria’s Brady bonds.



ferent groups of private and official creditors) is itself a daunting task.However, ranking explicit claims is not enough. There would still be aneed to figure out the relative seniority of explicit and implicit debt—suchas pension claims and guaranteed deposits—as well as the priority to begiven to paying debt relative to supporting government spending and

keeping taxes from rising. Specifying in advance the right balance be-tween increasing taxes, reducing government spending, reducing currentand future pension liabilities, imposing losses on depositors, reducing do-mestic debt, reducing external debt held by private creditors, and reduc-ing external debt held by official—bilateral and multilateral—creditors isnext to impossible.

The existence of so many creditors, claimants, and stakeholders makesthe resolution of any sovereign debt and financial crisis eminently a po-litical process that does not lend itself to simple legal rules. This is partic-

ularly true for ranking the sovereign’s domestic liabilities, which, as dis-cussed, can involve choosing between paying police salaries and paying bank depositors.

Is a Seniority System Enforceable?

A related issue is the enforceability of any formal seniority scheme. In thecorporate context, a bankruptcy judge can use the power of the state toprevent discriminatory payments to junior claimants and can exerciseconsiderable control over the distressed debtor’s assets. No one has simi-lar power over a sovereign in distress. As discussed in the next chapter,the difficulty in initiating successful litigation against a sovereign pro-vides it with substantial freedom of action. This potentially includes thefreedom to ignore a formal seniority scheme. The difficulties in enforcinga formal system of debt seniority are particularly acute with respect to do-mestic debts.

The inability to liquidate a sovereign and distribute the proceeds fromliquidation according to defined rules of absolute priority is not the crucial barrier to the creation of a system of sovereign priorities. The bigger bar-rier is the absence of a supranational power that can enforce a system of relative priorities (Gelpern 2004). Domestic bankruptcy law can set out apriority structure for semisovereign public entities, including subnationalgovernments that, like a sovereign, cannot be liquidated. US law allowscreditors to seek a court order forcing a municipal debtor to increase taxesto fund debt service, even if this has been very hard to do in practice.However, there is a difference between a municipal debtor that is subjectto national law and a sovereign.40 Sovereign countries have a degree of in-dependence that exceeds the legal independence of US states, and USstates, unlike US municipalities, are not subject to federal bankruptcy law.


40. Black’s Law Dictionary (6th ed., 1990) defines sovereignty as a “supreme, absolute anduncontrollable power by which any independent state is governed.”



A sovereign not subject to external enforcement might nonetheless con-clude that following an established system of priorities is in its own inter-est. Sovereigns do care about their reputation and failing to follow a gen-erally accepted seniority system would likely damage their future abilityto borrow. Sovereigns, for example, have found it in their interest to pay

the IMF and World Bank even when this has required sacrifices, in part be-cause of the reputational costs of trying to change the “rules of the game.”If the formal ranking were enshrined in national laws, then a nationalcourt might be able to issue an enforceable legal order that stopped dis-criminatory payments to junior creditors if more senior creditors were not being paid. Of course, such a remedy would be effective only if the sover-eign were making payments to more junior domestic debt holders. Seniorcreditors have few remedies if the debtor simply refuses to pay anyone. Fi-nally, official creditors that currently lend to a sovereign in distress on the

strength of their preferred status—notably the IMF—could use the lever-age that comes from their new lending to strengthen incentives to respecta defined system of priorities. The IMF might be willing to lend into ar-rears if the country had defaulted on its debts but not if the country also“defaulted” on the established system of priorities.

Proposals for a More Defined System of Sovereign Priorities

Most proposals to make the priority given to different sovereign debtsmore explicit set aside the relative treatment of external and domesticdebt and of private and official debt and instead focus on the creation of an explicit seniority regime for the sovereign’s private external debt. Thisfocus makes it much easier to define and enforce a clear system of priori-ties even as the more limited coverage limits the impact the new systemof priorities would have on the restructuring process. A number of thesesuggestions warrant further examination.

Bolton-Skeel Proposal: First to Lend, First to Be Paid

Patrick Bolton and David Skeel (2003) have proposed the creation of asimple system of seniority where debt issued earlier would be senior todebt issued later. This “first-in-time” priority system would eliminate theincentives sovereigns have to dilute their capacity to pay existing debt bytaking on additional debt. Such a proposal would no doubt achieve its in-tended objective of hindering overborrowing.

However, this particular proposal has shortcomings. First, it wouldfundamentally change the way sovereign debt trades in the secondary

market. The date of issuance would matter more than the bond’s residualmaturity. This means, for example, that a new five-year bond would beworth less (because of its lower position in the pecking order) than a ten-year bond that was issued five years ago and that has a residual maturity




of five years. This would make constructing a sovereign yield curve diffi-cult, since the standard yield curve implicitly assumes that all bonds thatmake up the curve have the same priority. Other technical issues to con-sider include the treatment of debt issued in voluntary debt exchanges foroutstanding long-maturity debt. No doubt the markets could adjust to

these new rules, but it may not be easy.Second, junior creditors who are lending at a later point in time would

have a strong incentive to find other ways of protecting themselves fromlosses. Markets adapt (Lipsworth and Nystedt 2001). Junior creditors mayinsist that their claims carry a short maturity. A first-in-time priority rulerisks creating a bifurcated sovereign debt market, with some long-termsenior debt that benefits from formal priority, some very short-term juniordebt, and perhaps very little in between. Consequently, giving formal se-niority risks creating incentives to increase the use of forms of debt that

make rollover/liquidity crisis more likely.Third, making the new system work would likely require additional

rules. The first-in-time priority rule would have exceptions. For example,as will be discussed in detail in the next chapter, most bankruptcy regimesgranted seniority to new financing to facilitate the operational and finan-cial adjustments required by a distressed debtor that is undergoing a fi-nancial reorganization. Since the IMF performs an analogous role in thesovereign context, presumably it would continue to receive some form of priority. Indeed, the denial of IMF financing would likely be central to the

enforcement of priorities in a world of sovereign states.Moreover, other rules may be needed to protect the intent of the prior-

ity structure. For example, some bankruptcy regimes require that firms inthe “zone of insolvency” be managed to protect the interest of senior cred-itors and even give them a claim on payments made to more junior cred-itors just before a formal bankruptcy declaration. The sovereign world hasa similar problem. A country that runs down its reserves to pay short-maturity junior debt just before default effectively damages—by deplet-ing its reserves—the interest of its long-term creditors in the process.

However, so long as the sovereign has not missed a payment before for-mal default, the formal seniority granted to long-term debt issued far inthe past offers creditors no protection against this risk. Taking steps tolimit the risk of the sovereign acting against the interest of its senior, long-term creditors could require introducing an additional layer of complex-ity. A regime that did not limit the risk of a sovereign fully depleting itsreserves by making payments to junior creditors before default would stillwork but with less effective protection for senior creditors.41


41. The same basic argument could be extended to other uses of the sovereign’s reserves. In-tervention in the foreign exchange market, for example, could deplete sovereign reserves be-fore default, damaging the interest of long-term creditors—as could lender-of-last-resortlending to the domestic banking system facing a run out of dollar deposits.



Finally, a first-in-time rule makes it cheaper for a policymaker that in-herits a low debt load to engage in a borrowing binge.42 Consequently, apolicymaker with a bias toward fiscal irresponsibility would have a strongincentive to issue as much cheap debt as possible early on. (Bolton andSkeel would give all debt issued in a given year the same priority.) At the

limit, overborrowing would be accelerated rather than reduced. Problemswith time consistency are an even bigger obstacle for other proposals tolimit sovereign overborrowing. For example, some have suggested that asovereign should commit to an upper limit on its debt stock (either ab-solute or a share of GDP) to lower borrowing costs. If the promise not to borrow above a limit actually lowered current borrowing costs, it might just encourage those policymakers to borrow more upfront (Zettelmeyer2003). Realistically, though, there is no way to prevent later policymakersfrom exceeding the proposed limit.

Soros Proposal: International Deposit Insurance Agency

Other proposals would provide seniority to part of the external debtissued by a sovereign in order to limit the risk of a liquidity crisis. Forexample, George Soros (1998; “Avoiding a Breakdown: Asia’s Crisis De-mands a Rethink of International Regulation,” Financial Times, December3, 1997) proposed the creation of an international deposit insurance agencythat would insure international investors’ claims against default. Sorosaims to reduce the risk of investor panics generating self-fulfilling crisesof confidence. Fully insured claims would not have any incentive to run.To reduce the risk of moral hazard, Soros suggests that the amount of in-surable claims be capped. The IMF would set the ceiling on insured bor-rowing, and debtor countries would pay the cost of this insurance schemethrough an insurance fee.

This proposal has a number of problems.43 If the insurance fee is actu-arially fair and there are no informational failures, then the cost of issuinginsured debt would not differ from that of issuing uninsured debt. A fullyinsured bond would be riskless and have no spread relative to other risk-less international bonds. However, the insurance fee should be equal tothe spread of that country’s uninsured debt. Including the insurance fee,the cost of external borrowing for the country would remain the same.44


42. In the Alesina and Tabellini (1990) model of excessive deficits and debt, for example, thesovereign overborrows because a government that may not be in power in the future dis-counts the welfare of future taxpayers/generations too much.

43. See also Eichengreen (1999) for a thoughtful discussion.

44. If market prices are different from actuarially fair prices, then the debtor may gain orlose. Differences in the relative knowledge or ignorance about fundamental risks betweenthe debtor and market providers of insurance may be a channel through which value is cre-ated. But one has to rely on externalities or informational failure to make such an argument.



Of course, setting the right fee structure would be difficult, and therecould well be an implicit subsidy. The IMF also would have to protectagainst the risk that politically important countries would use their lever-age to obtain a higher borrowing limit than their fundamentals warrant.45

The desire to limit the risk of runs motivated Soros’ proposal for the in-

ternational equivalent of deposit insurance. However, to truly eliminatethe risk of runs, a debtor would have to be precluded from issuing unin-sured debt in excess of the cap that was set on its insured debt. Uninsuredclaims would continue to have strong incentives to run.46 Such debt might be more expensive to issue, and thus countries may issue less of it. Cred-itors, however, also would have strong incentives to make sure that theyhad the option to exit from such “uninsured” lending quickly and thus tolend only for relatively short terms. There is no guarantee that the risk of liquidity runs would be reduced—insured claims would not run, but

uninsured claims would have strong incentives to get out quickly.Soros suggested that one benefit of his proposal is that the explicit in-

surance provided to some claims would make it easier to deny a bailoutto those claims that lacked insurance. However, this is not obvious. If let-ting the country default on its uninsured claims risked a deep crisis in apolitically important ally—or risked triggering contagion to other coun-tries that had issued large amounts of uninsured debt—in practice, theIMF or the G-7 might step in and bail out a country having difficultieswith its uninsured debt.

The Soros proposal does try to provide seniority in a clear and trans-parent way and to charge a fair fee for the provision of insurance. Manyother proposals to provide sovereign debt with some form of senioritythrough credit enhancements (collateral, guarantees, or other forms of in-surance) are less transparent. These proposals are usually motivated by adesire to help a sovereign that is having difficulty placing unsecured ex-ternal bonds—whether because of a run, growing doubts about its funda-mentals, or unsettled global market conditions—raise additional externalfinancing without generating the appearance of a bailout. They often use

partial credit guarantees from the World Bank or other MDBs to engineera class of semipreferred sovereign debt. These proposals, discussed in


45. A number of other difficult questions would also need to be addressed. For example, therolloff of cross-border bank lines has been a major source of pressure on the sovereign’s ownreserves, as most sovereigns step in and protect the banks during a crisis. If the deposit in-surance applied only to the sovereign’s external borrowing, it would not stop all runs. If pri-vate banks were eligible, the insured loan would either need to be explicitly guaranteed bythe sovereign, which would assume responsibility for supervising the bank, or the interna-tional deposit insurance agency would need to assume some regulatory functions.

46. This shortcoming is common to all proposals that guarantee (or provide collateral for)some claims on a sovereign but not others: Claims that are not guaranteed or enhanced stillhave an incentive to run.



flight. The priority structures of US states have not been tested by an actualrestructuring—and may never be. The US government would probably notlet a state like California default.

Incentives for a country to follow its own system of priorities could bestrengthened by denying IMF lending to a country that does not follow its

own system. But IMF enforcement has its own difficulties. The IMF’s ownprior exposure constrains its leverage—denying IMF financing to a sover-eign that has defaulted on its own priority structure could result in a de-fault on the IMF. Even setting aside the IMF’s interests as a creditor, it may be better off using its limited leverage to improve a country’s macroeco-nomic policy framework rather than forcing a country to stick to its prior-ity structure. The details of the priority structure will likely be contested,and the IMF may even conclude, for example, that protecting the coun-try’s banking system is more important than religiously following any

preexisting priority structure. The international community’s interest, writ broadly, would be better served by helping the country recover ratherthan punish it for failing to live up to its self-defined system of priorities.47

Still, Gelpern’s proposal has greater merit and fewer costs than propos-als for more radical reform. It could force a sovereign to make its priori-ties in default explicit, and once priorities are made explicit, the reputa-tional cost of defaulting on an articulated priority structure would beincreased. Too many high-risk sovereigns prefer to think that default isunthinkable and that no contingency plan for a default is needed. Yet even

if sovereigns are expected to lay out its intended system of priorities, therisk of a sovereign concluding that it is not in its interest to follow its self-chosen priority regime is real.

Conclusion

The absence of well-defined rules of priority clearly is a source of uncer-tainty for investors following a sovereign default. Sovereigns have sub-

stantial—though certainly not unlimited—freedom to determine ex postthe priority structure that is in their interest. Investors who base invest-ment decisions on the types of debt instruments that were favored in theprevious restructurings will often be disappointed. Since holding patternsvary, the type of instruments that get relative priority also tends to vary.Look at the difference between Russia and Argentina’s external bonds.Russia paid external bonds issued after the fall of the Soviet Union in fulland on time even as it nearly wiped out the real value of its domestic debtand obtained significant concessions from holders of its Soviet-era exter-


47. Gelpern (2004) in the end recommends only that sovereigns be encouraged to lay outtheir own priority structures and that the IMF report on the sovereigns’ adherence to theirself-selected structures as part of its surveillance process.



nal debt. Argentina more or less has proposed doing the opposite. It isseeking a larger haircut from its external bondholders than it imposed—through a series of restructurings—on the domestic-law debt held by its banking system.

The core question in the debate on sovereign priorities is whether it is

possible to develop a system that both can be enforced and substantiallyimproves on the status quo. We are skeptical.

The current system has its own logic and its own norms. Experiencedmarket participants know that the IFIs have effective priority as a resultof the IFIs’ role as providers of new money in a crisis and that this prior-ity is sustained in part because the IFIs are typically willing to roll overtheir positions for some time. Savvy market participants also know that asovereign is likely to go into arrears on Paris Club debt before it suspendspayments on its privately held debt, since the risk of enforcement actions

from Paris Club creditors is negligible. Long-term syndicated bank loansand international sovereign bonds are both likely to be restructured if thesovereign has significant amounts of bonds and loans outstanding. It ishard for a sovereign seeking deep debt reduction to exclude some of itsinternational bonds entirely from the restructuring—one bond issue hasno reason to agree to voluntarily subordinate itself to another bond issue.Claims to collateral are likely to be respected. Short-term external debt—including the claims of international banks on local private banks, whichare more often than not implicit liabilities of a sovereign—lacks formal

seniority but can often get out before the restructuring.The biggest uncertainties arise as a result of domestically held debt—in-

cluding foreign-law debt held domestically. The current rules of the gamereflect the self-interest of sovereign governments. A government in de-fault will almost always take action to limit the losses depositors bear. Thegymnastics that Argentina went through to create a subset of performingdomestic debt out of an initial debt stock that was heavy on international bonds are a case in point. Incentives to favor domestic debt are furtherstrengthened by an expectation that it will be easier to regain access to

domestic rather than international markets. A restructuring regime thatgrants the sovereign substantial discretion will almost always result inpreferential treatment of debt held by banks or in the creation of per-forming bank recapitalization bonds to limit the size of any haircut on de-positors. This is not bad: We have tried to show that treating the domes-tic debt restructuring as part of the broad domestic negotiation about howto apportion the burden of domestic adjustment has economic and politi-cal logic.

Even if these broad norms are followed, there is no shortage of residual

uncertainty. However, a much clearer priority regime may not be possibleto design or feasible to enforce, given the scale and complexity of the fi-nancial claims on a sovereign and the difficulties forcing a sovereign gov-




ernment to follow a priority structure, if it does not believe it reflects itsinterests. Even radical institutional change, such as the creation of an in-ternational bankruptcy regime, might not create a very precise system of sovereign priorities. It is hard to write rules sufficiently precise to avoidthe need for interpretation, or rules that creditors would not contest the

through the courts.48What really matters is whether relative priority can be assessed ex ante

with some reasonable degree of predictability and without an excessivelylarge number of surprises. The current nonsystem could be codified overtime. Yet even a real effort to make the existing rules of the game moreexplicit is unlikely to result in a true system of priorities that defines the rel-ative treatment of all financial claims on the sovereign. Sovereign commit-ments to constrain their own freedom of action are only credible to a point.The current regime has costs, but its current fuzziness and ambiguity also

may be inherent to any regime that deals with sovereign borrowers.


48. See Tarullo (2001) and Gelpern (2004) for the argument that any formal regime wouldneed the authority to make new rules.

senioirity of sovereign debts

Documents